The German Savings Puzzle

Transcription

The German Savings Puzzle
Axel Börsch-Supan
06-2002
January 2002
mea – Mannheimer Forschungsinstitut Ökonomie und Demographischer Wandel
Gebäude L 13, 17_D-68131 Mannheim_Sekretariat +49 621 181-2773/1862_Telefax +49 621 181-1863_www.mea.uni-mannheim.de
International Comparison of
Household Savings Behaviour: The
German Savings Puzzle
International Comparison of Household Savings Behaviour:
The German Savings Puzzle
Content
This mea discussion paper presents excerpts of the International Savings Comparison Project
covering household savings behaviour in seven countries. The whole series of comparative
country studies can be found in a special issue of the journal “Research in Economics”,
Volume 55, Number 2, June 2001.
The introduction gives an outline of the research program of the project.
A project as complex as the International Savings Comparison Project has sparked
discussions and controversy. Tullio Jappelli, who has accompanied the International Savings
Comparison Project since its inception, has accepted the invitation to write a critical
discussion of the project’s work so far. His discussion and a brief rejoinder by Axel BörschSupan compose the second and third part of this discussion paper.
The last part of this paper presents the results of this project for Germany.
International Comparison of Household Savings Behaviour:
A Study of Life-Cycle Savings in Seven Countries
Introduction
By Axel Börsch-Supan1
1. Purpose of this special issue
Household saving is still little understood, and even the basic facts – for instance: How does
saving change over the life cycle? Do the elderly draw down their wealth? – are controversial.
Understanding saving behaviour is not only an important question because the division of
income in consumption and saving concerns one of the most fundamental household
decisions, but it is also of utmost policy relevance since private household saving as a private
insurance interacts with social policy as public insurance. Population ageing and its threat to
the sustainability of the public insurance systems has put the spotlight back on own saving as
a device for old-age provision. Solving the pension crises therefore requires understanding
saving.
This special issue of Research in Economics is devoted to a further step in this direction. It
presents a first stock taking of the International Savings Comparison Project – a project
performed under the auspices of a European Union sponsored network of researchers. 2 The
main focus of this project is the interaction of household saving with public policy, notably
the generosity and design of public pension systems. It is very much in the tradition of
Feldstein’s (1974) seminal study, but we transpose the inference from time series data to a set
of international panel data.
Our inference is based on seven country studies. For space reasons, five country studies
appear in this issue, while two country studies will appear in the following issue. The
countries range from five European countries (France, Germany, Italy, the Netherlands and
the United Kingdom) to Japan and the United States. In all these countries, pension reform is
high up on the policy agenda. All countries have already introduced, or are contemplating
introducing, the augmentation of their pay-as-you-go public pension systems with private
(occupational and individual) funded pension plans. For this reason alone, the seven countries
1
2
I am grateful for comments by Agar Brugiavini, Tullio Jappelli, Guglielmo Weber, and Joachim Winter.
The TMR (Training and Mobility of Researchers’) network on “Savings, Pensions and Portfolio Choice”.
1
are interesting subjects for a study of saving behaviour.
The combination of the seven country studies, however, should be more than the sum of its
parts. Understanding saving behaviour requires variation in the potential determinants of
saving. Studies within a single country, however, often lack the necessary variation in public
policies: the counterfactual is missing. This is most germane for cross sectional data from a
single country that usually fail to have any policy variation. Traditional studies of household
saving have therefore exploited the time series variation in aggregate data. Such studies,
however, cannot really account for changes in the composition of a heterogeneous population.
One obvious solution is to use panel data. Panel data sets that contain saving data, however,
are usually short and therefore rarely include policy changes and “historical experiments”.
This particularly applies to our main determinant of interest: pension policy is not changed
frequently, and this for a good reason. The main idea behind the International Savings
Comparison Project is to exploit international variation that might provide additional variation
in policy variables because different countries have widely different social policies, capital
taxation regimes, etc.
A first objective of this project is therefore to set up a comparable longitudinal database that
permits more insight in the relation between pension and other policies on the one hand and
saving behaviour on the other hand. As a second objective, we test our working hypothesis: a
major part of the differences in the age-saving patterns observed across European countries,
Japan and the U.S. are generated by differences in national pension policies. The ultimate
objective of the line of research pursued in this project is to construct a model that predicts
life cycle saving patterns as a function of pension policies, taxes rules, and other determinants.
While this last objective is not a realistic goal for this special issue, our work will be guided
by such a frame of thinking.
More modestly, the papers fulfil two tasks. The first is descriptive: the papers collect the main
saving measures by age and cohort. The second task is interpretation: the papers link saving
patterns to country-specific policies, most prominently (but not exclusively) pension policies.
At this stage of the project, this link is a rather informal one.
Specifically, each paper focuses on two issues:
• To measure how saving changes during the life cycle. This requires the separation of age
and cohort effects, subject to a common treatment of time effects. It also requires a
common definition of saving components in the various countries.
• To augment saving data by data on pensions. This includes mandatory contributions to
unfunded pension plans on the one hand, and data on retirement income by source on the
2
other hand.
The work is in the tradition of earlier cross-national studies, and we are happy to be able to
leverage earlier work – often done by the authors themselves – to new connections and
insights. A particularly noteworthy foundation are the age-saving profiles for the G-7
countries (except France) that have been presented in the volume edited by Poterba (1994),
referring to saving data until the mid 1980s. We update these profiles and base them on
stricter common definitions. In addition, we extend the Poterba volume, which mostly relied
on cross sectional evidence, by purging the age-saving profiles from cohort effects drawn
from longitudinal data. This is important because apparent life-cycle effects in cross sectional
data are severely confounded by changes from cohorts to cohort. How severe the resulting
bias can be is demonstrated further below.
The papers in this special issue are brief and concise. They are, as mentioned before, only a
first stock taking which stresses the main features in each country. The International Savings
Comparisons Project will proceed with a second set of studies that are more detailed and more
tightly structured around a set of common descriptions and analyses. The reader is referred to
these papers which will appear in a volume (Börsch-Supan, 2001). This volume will also
provide an extensive discussion of the methodological issues in identifying and measuring
savings (Brugiavini and Weber, 2001) to which we only briefly allude in this introduction.
Moreover, the volume will include a machine-readable appendix with the underlying data that
will enable readers to generate alternative specifications of saving aggregates and to apply
alternative assumptions for the separation of age, cohort, and time effects in saving behaviour.
2. Methodology
The papers in this special issue use a set of common saving concepts that are defined in the
first part of this section. While these accounting definitions are tedious and may not be the
matter that excites most economists, they are a crucial necessity for a meaningful crossnational comparison. The second part describes our approach to separate age, time, and cohort
effects – a crucial requirement to analyse saving over the life course. As mentioned, a more
extensive discussion of the various approaches to measure and identify saving behaviour is
provided by Brugiavini and Weber (2001).
2.1 Saving Concepts
The starting point for our various saving concepts is a macroeconomic point of view: saving
is the addition to the physical capital stock, Wt, during the period from time t-1 to time t. The
central underlying equation is
3
(1)
Wt = (1+rt) Wt-1 + Yt - Ct
where Yt stands for disposable labour and transfer income, net of taxes and contributions to
unfunded social security schemes, and Ct for consumption expenditures. We will come back
to this disposable income definition later. Capital income is rtWt-1 for the rate of return rt.
We first distinguish between active and passive saving. Passive saving are capital gains that
are automatically reinvested – the most salient example is stock market appreciation. If all
capital income is automatically reinvested, and let us assume this for the rest of this
introduction, active saving in equation 1 is (Yt -Ct) while passive saving is rtWt-1.
With suitable data, saving can be disaggregated in its portfolio components. Ideally, we
observe daily inflows into, and outflows from, each separate account. We denote the active
part of these in and outflows by Dit such that the sum over all portfolio items i yields (Yt -Ct).
Hence,
(2)
Wit = (1+rit) Wit-1 + Dit
where Wit and rit denote the respective stocks and returns.
Of particular importance for our analysis of household saving behaviour is the distinction
between discretionary and mandatory saving. Discretionary saving is completely under the
control of the households. The households choose its absolute value as well as its portfolio
composition, given their budget constraints and applicable incentives such as tax relief and
mandatory contributions to funded and unfunded pension schemes. In turn, mandatory saving
is beyond the control of the household. The most important example is mandatory
contributions to funded occupational pension plans. Here, the volume is prescribed (e.g., a
fixed absolute sum or a fixed percentage of gross income) and frequently even the portfolio
composition is outside the control of the household (e.g., the employer provides a single
pension plan).
Where possible, we also distinguish between financial and real saving. This is now a
microeconomic concept, departing from the macroeconomic view that all saving will
ultimately be physical saving. Active discretionary financial saving is:
•
Deposits into, minus withdrawals from, saving accounts, mutual money market accounts,
and other money-like investments
•
plus purchases of, minus sales of, bonds
•
plus purchases of, minus sales of, stocks
•
plus contributions to, minus out payments from, whole life insurance
•
plus contributions to, minus out payments from, dedicated saving plans (defined by a
contract that determines for which purpose withdrawals may be made, e.g., building
4
societies, individual health spending accounts, etc.)
•
plus voluntary contributions to, minus payments from, individual retirement accounts and
pension funds where withdrawals may be made only after retirement or a prespecified age
•
plus amortisation of, minus take-up of, consumer loans.
In turn, active discretionary real saving consists of:
•
Purchases of, minus sales of, real estate (including owner-occupied housing)
•
plus expenditures in upkeep and improvement of housing, minus 2% depreciation
•
plus amortisation of, minus take-up of, mortgages
•
plus purchases of, minus sales of, gold and other jewellery.
We also report the corresponding stock measures, financial and real wealth. Note that
mortgage loans count as (negative) real wealth.
We started by defining saving as additions to the physical capital stock. Some economists,
however, prefer a broader definition of saving that also includes the addition of claims on
unfunded pension benefits (Jappelli and Modigliani, 1998). Under such a broad view,
contributions to pay-as-you-go financed pension schemes are considered saving. We will use
at times the term “notional saving” for these contributions although we are aware that the
term “saving” may be confusing here since these contributions do not contribute to the capital
stock. Consequentially, receiving pension benefits is “notional dissaving”, and the current
present value of pension benefit claims is dubbed “notional wealth”, “unfunded pension
wealth”, or “social security wealth”.
While it may be controversial whether it makes semantic sense to call contributions to pay-asyou-go systems “saving”, it is uncontroversial that it is important to take account of these
contributions because they may substitute for actual saving. As a matter of fact, it is just this
potential substitution which is at the core of this project and the papers in this special issue.
Similar to equation 1, the stock of social security wealth, SSWt, evolves from time t-1 to time t
by active contributions, Tt (negative: benefit receipts, Bt) and passive appreciation at the
pension systems internal rate of return, irt:
(3)
SSWt = (1+irt) SSWt-1 + Tt - Bt
Jappelli and Modigliani (1998) combine physical wealth Wt with notional social security
wealth SSWt to a measure of “total wealth” TWt. By defining earned income as
(4)
YEt = Yt + Tt - Bt
which is gross labour income net of taxes (but not net of social security contributions on the
one hand, and not including transfer income from the social security system on the other
5
hand), one can combine equations (1) and (2):
(5)
TWt = (1+r*t) TWt-1 + YEt - Ct
where r* is the implied return on TW. While it is tempting to construct such a measure of
“total wealth”, we will not pursue this avenue because we think that physical and notional
wealth are very different concepts in the minds of most households. One is bequeathable, the
other not. Physical wealth can be borrowed against, which is not possible for “notional
wealth”. We also need strict assumptions on the time evolution of the two rates of return to
consistently aggregate them into r*. Hence, combining the two will lead to an, in our view,
unacceptable loss of information.
Instead, the papers in this special issue will compute a simple measure of social security
wealth based on equation 3. By using equilibrium forecasts of Tt and Bt, usually provided by
each country’s social security administration, and assuming a zero internal rate of return, the
papers will report the accumulation of claims to pension benefits up to the normal retirement
age, and then show its “notional decumulation.” The assumption of a zero internal rate of
return is chosen mainly for convenience; it may, however, not be too bad an approximation to
future returns of most pay-as-you-go pension systems due to population ageing.
Let us now return to saving in the narrower sense. Equations 1 and 2 show that, at least in
principle, there are three different ways of measuring (physical) saving:
♦ first, by comparing asset holdings at the beginning and at the end of a period: Wt - Wt-1
♦ second, by adding inflows and outflows of wealth accounts during one year: Σi(ritWit-1+Dit)
♦ third, by taking the residual of income minus consumption expenditures: (Yt+rtWt-1) - Ct
Equality of these measurement concepts is only achieved when the variables involved – stock
of wealth, flows into and out of accounts, income and expenditures – are consistently defined.
Part of the exercise in this special issue is to achieve such internal consistency.
Ideally, we would like to report all three measures in order to learn how reliable actual
measurements of Wt, Dit, Yt and Ct are. In practice, however, the data sources available to the
seven country studies are less than satisfactory. In many countries, only two measures can be
computed, in some countries only one. Frequently, capital gains are not measured or have to
be imputed using aggregate data on rates of return that is likely to produce major
measurement errors in particular for highly localised real estate. Moreover, the distinction
between discretionary and mandatory saving can only be made when we have a detailed
account how total saving is split among different usages. This is obviously not possible for the
6
residual saving measure (when consumption is subtracted from income). One of the main
lessons of this special issue is that research on saving behaviour is still severely hampered by
the lack of suitable data. This is astounding since pension reform, an important policy issue in
all of the seven countries studied, requires a thorough understanding of the substitutability
between discretionary and mandatory saving as well as between physical and notional saving.
The papers will show that we still only rudimentarily understand these substitution effects.
2.2 Construction of longitudinal data and identification
Saving behaviour will not only change by age, as the life-cycle theory predicts, but also from
cohort to cohort. Younger cohorts have experienced peace and stability, while the cohorts that
are now in retirement have lived through one or even two World Wars and the Great
Depression. In addition, household saving will react to the business cycle and similar factors
at any given point in time. In this section, we briefly discuss the simple methodology by
which the papers in this special issue separate age, cohort and calendar-time effects from each
other.
In cross sectional data, each age category also represents a cohort. Thus, we cannot
distinguish between cohort and age effects. Moreover and trivially, a single cross section
cannot identify the effects of calendar-time specific events since we only observe one single
point in time. Figure 1 shows the errors one makes when ignoring this first fundamental
identification problem. It is taken from the German country study. Comparing points on one
of the cross sectional lines drawn in Figure 1 does not depict life-cycle changes since one
compares households that are simultaneously in different age categories and cohorts. Hence,
the apparent hump shape of wealth in Figure 1 is a combination of age and cohort effects, not
the life cycle change created by age.
7
Figure 1: Wealth by Cross Section
Total wealth in DM
350.000
300.000
250.000
200.000
1993
150.000
1988
100.000
1983
50.000
1978
0
25 30 35 40 45 50 55 60 65 70 75 80 85
Age-Groups
Notes and Source: See chapter on Germany in this special issue. All amounts in 1993 DM. 1 DM in 1993
corresponds to a purchasing power of 0,57 Euro in 1999.
The main point in the analysis of this special issue is therefore to use longitudinal data to shed
more light on age and cohort effects. Unfortunately, only a few countries have panel data that
permit following an individual household over time. In most countries in this project, we only
have several unlinked cross-sections, such as the four cross sections displayed in Figure 1. We
do not even know whether a household has participated in two or more of these cross
sectional surveys because the identification of this household is impossible.
We therefore resort to the construction of synthetic panels. Households of each survey
(“wave”) are divided up into as many homogenous household types ("cells") as possible.
Next, these cells are identified across time. Such a panel does not consist of households but
household types as survey units. On the one hand, the statistical analysis of such synthetic
panels is eased by reducing the unobserved heterogeneity by taking means within household
types. On the other hand, as Deaton (1985) has analysed, neglecting movements between
household types across time may lead to biases. As long as there is no panel of individuals
with savings data, we will have to live with a conflict between the stability and homogeneity
of cells.
8
There are also more mundane problems with synthetic panels. In order to obtain consistent
variable definitions across waves, one has to take into account the differences between
surveys. Sometimes newer waves contain more detailed information than earlier ones, and
frequently variable definitions change from survey to survey. The studies in this special issue
must make various compromises between full comparability and best usage of available
information.
While cross sectional data identify only one dimension, as shown in Figure 1, longitudinal
data permit the identification of two dimensions. Since there are, however, three effects – age,
cohort, and calendar-time effects – we are still stuck with a fundamental identification
problem because these three effects are strictly collinear, calendar time being the sum of birth
date and age.
Only strong assumptions can therefore identify life cycle saving patterns. One assumption,
that is as simple as brutal, is to subsume time effects – by setting them to zero – into age and
cohort effects. This is the method applied to the papers in this special issue. Departing from a
set of cross sections, such as those depicted in Figure 1, we identify households in subsequent
five-year age-groups with each other, i.e., by identifying the 45-49 year old persons in 1978
with the 50-54 year old persons in 1983, the 55-59 year old persons in 1988, and the 60-64
year old persons in 1993. This procedure amounts to re-connect the points of Figure 1 in a
different fashion, see Figure 2:
Figure 2: Age and Cohort Effects if Time Effects are Zero
Total wealth in DM
350.000
Cohort born in 1933
300.000
Cohort born in 1923
250.000
Cohort born in 1918
200.000
1993
150.000
1988
100.000
1983
50.000
1978
0
25
30
35
40
45
50 55
60
65
70
75
80
85
Age-Groups
Notes and Sources: See chapter on Germany in this special issue.
9
We then redraw the dotted lines in Figure 2 to display the age-profiles of wealth by cohort,
see Figure 3, starting at the left side with the youngest cohort in our data, born between 1954
and 1958, and proceeding to the oldest cohort, born between 1909 and 1913.
Figure 3: Wealth by Cohort
350.000
Kohorten
300.000
1909
total wealth
1914
250.000
1919
200.000
1924
1929
150.000
1934
1939
1944
100.000
1949
50.000
1954
0
20
25
30
35
40
45
50
55
60
65
70
75
80
85
age group
Notes and Sources: See chapter on Germany in this special issue.
Note that the cohort-corrected profiles in Figure 3 leave nothing from the apparent hump
shape that was suggested by the cross sectional Figure 1: all age profiles monotonically
increase with age, except a little blip among the very old.
Figure 4 applies the same technique to saving rates, the variable in the centre of our interest:
10
Figure 4: Median Saving Rates by Cohort
16%
14%
12%
1914
Saving Rate
10%
1919
1924
1929
8%
1934
1939
1944
6%
1949
4%
2%
0%
25
30
35
40
45
50
55
60
65
70
75
80
85
Age-group
Notes and Sources: See chapter on Germany in this special issue.
In order to smooth these ragged profiles, we regress the observed saving rates on a (fifthorder) polynomial in age and a (third-order) polynomial in cohort (alternatively: a set of
cohort indicators). This leads to Figure 5. In this figure, age and cohort effects are much more
clearly visible than in Figure 4. We can dissect in the profiles in Figure 5 into a “pure” age
and a “pure” cohort effect, see Figure 6. Note the quotation marks: these effects are “pure”
only insofar as they crucially depend on our identifying assumption of zero time effects. The
left panel of Figure 6 emerges from Figure 5 when the intercepts of each segment (the “pure”
cohort effects) are set to a common value – the “pure” age or life-cycle effect remains; the
right panel of Figure shows the evolution of the intercepts from cohort to cohort.
11
Figure 5: Smoothed Saving Rates by Cohort
16
14
12
10
8
6
4
2
0
20
23
26 29
32 35
38 41
44 47
50
53 56
59 62
65 68
71 74
77
80
Source: Same data as Figure 4, see chapter on Germany in this special issue.
Figure 6: “Pure” Age and Cohort Effects
cohort effect
age effect
0
19
7
0
19
6
0
0
19
5
19
0
80
75
70
65
60
55
50
45
40
35
30
25
20
0
0
0
2
19
4
4
19
3
6
0
8
19
2
10
19
1
12
0
18
16
14
12
10
8
6
4
2
0
14
Source: Slope and intercepts from Figure 5.
There are other and more sophisticated methods to separate age, cohort, and calendar-time
specific effects. One approach is to rescue at least the essence of time effects by not setting
them to zero as in the above-mentioned regression, but to include time dummies and to
impose the restriction that they sum to zero and are orthogonal to a linear trend (Attanasio,
1994; Deaton and Paxson, 1994). Yet another approach is to try to break the correlation
altogether by parametrizing the calendar-time specific effects. Alessie, Kapteyn and Lusardi
(1998) have pursued this line of identification and used a parametric function of productivity
growth and social security benefits to represent the effects of calendar time. The reader is
referred to Brugiavini and Weber (2001) for a more in-depth discussion of these identification
strategies and their advantages and disadvantages.
12
3. First results and a tempting interpretation
Figure 7 presents saving rates in three of the seven countries in this special issue: Germany,
Italy, and the Netherlands. The figures show the fitted values by age in each observed year
together with the upper and the lower point of a 95% confidence interval. They therefore offer
a visual impression of the stability and precision of these age profiles – stability in terms of
changes from year to year, and precision in terms of estimated standard deviations.
Figure 7: Cohort-corrected saving rates by age (medians)
Germany
Fitted values
pslo
Italy
Fitted values
pslo
pshi
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
The Netherlands
pshi
Fitted values
pslo
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
15
14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
20
30
40
50
age
60
70
80
pshi
20
30
GE: savrat by age
40
50
age
IT: savrat by age
60
70
80
20
30
40
50
age
60
70
80
NL: savrat by age
Sources: See country chapters in this special issue.
Figure 7 gives a good impression of the diversity of age-saving profiles. First, levels are very
different: Italian households have a high saving rate that exceeds the age-specific saving rates
of Dutch households at all ages, with German households in between. Second, Italian
households have experienced quite different saving rates over time (and possibly cohort)
while the savings behaviour of Dutch and German households was much more stable. Third,
the saving rates of Dutch and Italian households can be fairly precisely measured in a
statistical sense. This is also true for German households who are in the middle age groups
while sampling errors are large for young and elderly households. Fourth and finally, the lifecycle patterns are rather different: In Italy, a decline in saving rates comes late (after age 60).
In Germany, households save less after about age 40 but savings rates appear to slightly (not
significantly) increase at old age again (see also Figures 4 through 6). The median elderly
household in Germany and Italy does not dissave – in Germany, the saving rate stabilises at
around 4% and in Italy it remains even higher also in old age. This is quite different in the
Netherlands where the median saving rate is about zero for elderly households and slightly
negative for the oldest old.
What explains these startling differences? The honest answer is that we still do not know. It is
tempting, however, to consider the pension systems in those three countries as a working
13
explanation. Germany and Italy have pay-as-you-go financed public pensions with very high
replacement rates. They generate net retirement incomes that are approximately 70% of preretirement net earnings in Germany and may even exceed 100% in Italy.3 In addition, the
public pension systems in Germany and Italy provide generous survivor benefits that
constitute a substantial proportion of total unfunded pension wealth, and disability benefits at
similar and often even higher replacement levels than old-age pensions. As a result, public
pensions are by far the largest pillar of retirement income in these countries and constitute
more than 80% of the income of households headed by persons aged 65 and older, while
funded retirement income, such as asset income from private saving or firm pensions in which
the employer saves on behalf of the worker, plays a much smaller role. This is quite different
from the Netherlands which only provides a flat base pension on a pay-as-you-go basis with a
replacement rate that is very low for households above median income. All other retirement
income is withdrawals from mandatory occupational and individual pension accounts. Hence,
a crucial difference between the three countries in Figure 7 is that saving for old age is
unlikely to be the main savings motive in Germany and Italy, while it is necessary for Dutch
households. The famous hump shape of savings predicted by the life-cycle hypothesis
therefore applies to Dutch households, while (physical) savings are relatively flat in Germany
and Italy – in turn, “notional” social security wealth increases and decreases faster in
Germany and Italy than in the Netherlands, see the individual country studies.
If this explanation of the observed cross national saving differences were correct, it has
important implications for the future. If indeed most of the saving patterns currently observed
in Germany and Italy are caused by generous retirement benefits from their pay-as-you-go
pension systems, we should expect distinct changes in saving patterns when the pension
reforms in these countries will be put in place. The introduction of multi-pillar systems with a
substantial portion of funded retirement income will revive the retirement motive for saving.
In fact, these reformed systems will look very similar to the current Dutch system. Hence, it is
likely that saving rates among the young will increase (to accumulate retirement savings), and
saving rates among the elderly will decline sharply (because they will dissolve their
retirement savings).
So far for succumbing to the temptations of a monocausal interpretation. Unfortunately, life is
more complicated than permitting such simple inference – too many other factors, from real
estate prices through the organisation of financial markets, are likely to confound this
comparison. Much more research and much better data are needed to establish causality. The
3 See Gruber and Wise (1999) for a comparable description of the Dutch, German and Italian pension systems.
14
papers in this special issue are designed to water the readers’ mouth for such research, and
they should make the point that without proper longitudinal data on savings and wealth, we
will keep making pension policy without understanding the most basic behavioural effects of
such policy.
References:
Alessie R.J.M., A. Kapteyn, and A. Lusardi (1998), Explaining the Wealth Holdings of
Different Cohorts: Productivity Growth and Social Security, CentER Discussion
Paper, Tilburg University.
Attanasio, O.P. (1994), Personal Saving in the United States, in: J. Poterba (ed.), International
Comparisons of Household Savings, Chicago, London: University of Chicago Press,
57-124.
Börsch-Supan, A. (ed.) (2001), International Comparisons of Household Saving, New York:
Academic Press.
Brugiavini, A., and G. Weber (2001), Household Savings: Concepts and Measurement. In:
Börsch-Supan, A. (ed.), International Comparisons of Household Saving, New York:
Academic Press.
Deaton, A. (1985), Panel data from time-series of cross-sections. Journal of Econometrics,
30, 109-124.
Deaton, A. and C. Paxson (1994), Saving, Aging and Growth in Taiwan, in: D. Wise (ed),
Studies in the Economics of Aging, Chicago: The University of Chicago Press.
Gruber, J., and D. Wise (1999), Social Security and Retirement Around the World, Chicago:
The University of Chicago Press.
Jappelli, T., and F. Modigliani (1998), The Age-Saving Profile and the Life-Cycle
Hypothesis, CSEF Working Paper No. 4, University of Salerno.
Poterba, J. (ed.) (1994), International Comparisons of Household Savings, Chicago, London:
University of Chicago Press.
15
International Comparison of Household Savings Behavior:
A Study of Life-Cycle Savings in Seven Countries
Comment
by Tullio Jappelli
I thank Axel Borsch-Supan, Mario Padula and Luigi Pistaferri for useful discussions and
insights. The usual disclaimers apply.
16
The readers of the papers of the International Comparison Project of Household
Savings Behavior will immediately realize that the project addresses fundamental issues in the
economics of saving. The challenge of the project is to relate the international differences in
household saving to the variables that theory suggests should affect saving: demographics,
lifetime income profiles, taxation, access to credit markets, pension arrangements, social
insurance schemes, and other institutions affecting the decision to save. The studies focus
primarily on one crucial dimension of international differences, i.e. that different pension
arrangements should affect saving behavior.
The individual papers do not propose formal tests of this hypothesis. Rather, they aim
at summarizing the main facts about saving in each country and at describing the institutional
differences that are likely to affect household saving in each nation. In this respect, the project
follows a similar approach of an NBER project conducted few years ago (Poterba, 1994).
With the exception of France, that project centered on the same set of countries included in
this special issue of Research in Economics. The conclusion of the NBER project was that it
is very hard to account for differences in household saving rates across countries. Although
we have made important progress in terms of quality of data and measurement issues, in
summarizing the main findings of the project Borsch-Supan (2001) states that we still don’t
understand these international differences. It is therefore natural to ask what can we learn
from comparison studies of the sort performed by the International Project, and what are the
lessons for future research.
The life-cycle model inspires most of the country studies of this project. This model
posits that the main motivation for saving is to accumulate resources to be drained down for
later expenditure and in particular during retirement. According to the model, saving should
be positive for households in their working span and negative for the retired ones, so that
wealth should be hump-shaped (Modigliani, 1986). Mandatory contributions to pension funds
and to social security reduce the need to save for retirement (Feldstein, 1974). The idea is
therefore to exploit the international differences in pension arrangements to explain
differences in patterns of household saving. This approach is potentially very useful and
informative.
The main findings of the international project are summarized by Borsch-Supan (2001,
see especially Figure 7). In Italy, Germany, and the Netherlands saving rates are positive
through life. In France, the U.K. and the U.S. the qualitative pattern is similar: saving is
positive even in old age, and there is little evidence that the elderly decumulate wealth. Only
17
in Japan there is some evidence of wealth decumulation. 4 The project therefore challenges the
fundamental prediction of the life-cycle model. I will focus my comments around four main
issues: (1) the definition of saving adopted in most of the country studies; (2) measurement
issues; (3) the evidence; and (4) lessons for future research.
1. Saving definitions
The saving rates that are computed on microeconomic data for the 7 countries of the
International Comparison of Household Savings Behavior are for the most part based on a
concept of disposable income that does not take into account the role of mandated saving
through private and public pension systems. Disposable income treats contributions to
pension funds and to social security as taxes, and pension benefits as transfers. But since
contributions entitle the payer to receive a pension after retirement, they should be regarded as
a (compulsory) component of life cycle saving and hence added back to income. On the other
hand, income from pension funds and social security benefits accruing to the retired do not
represent income produced, but rather a drawing from the pension wealth accumulated up to
retirement. The greater the amount of mandatory saving, the greater is the difference between
earned income and disposable income.
• Jappelli and Modigliani (1998) argue that where mandatory contributions to
pension funds and social security are sizable (as in all countries of the International Project),
the age profile of saving as conventionally measured cannot be taken as evidence in favor or
against the life-cycle model. To illustrate this point and the relation between discretionary
saving and mandatory saving (or between discretionary wealth and pension wealth) it is useful
to start from the individual dynamic budget constraint. To make the argument in the simplest
possible way, assume that the real interest rate is zero and that the horizon is certain:
x = α+ p=w-c
4
The conclusions of the NBER project were similar. Poterba (1994) reports that in virtually all nations the
median saving rate is positive well beyond retirement, concluding that “the country studies provide very little
evidence that supports the life-cycle model”.
18
where x is total wealth, w earnings and c is consumption. Total wealth is the sum of
discretionary wealth a and pension wealth p. The difference between the two is that a is
bequetable while p is annuitized, and disappears when the individual dies. The right-hand-side
variables indicate that total saving (or change in total wealth) is the difference between
earnings and consumption.
Earned income (w) includes mandatory contributions to pension funds and social
security contributions (τ) and excludes pension benefits (b). From the standpoint of the lifecycle model, this is the relevant income measure. The conventional measure of disposable
income is instead y = w − τ + b . According to the two income definitions, one can compute
two measures of saving, discretionary and mandatory saving:
sd = α = w − τ + b − c
sm = p = τ − b
The former is the conventional measure of saving, the latter is the difference between
mandated pension contributions and benefits. Finally, total saving is the sum of discretionary
and mandatory saving sT = x = sd + sm .
To understand the importance of the different definitions, consider the case of a
worker with a constant age-earnings profile of 15,000 euros (before contributions) who starts
working at age 20, retires at age 60 and receives a pension until he or she is 75 years old. The
worker contributes a constant fraction of his earnings (25 percent) to a pension fund (or to the
social security administration) and receives a constant benefit of 66 percent of his pre-tax,
pre-retirement income, so that the present value of benefits equals the present value of
contributions (the pension is actuarially fair). In the example the replacement rate is quite
close to that observed in several of the countries of the international project (such as Italy,
Germany, the Netherlands). Suppose also that desired consumption is flat (at 10,900 euros),
and that the interest rate and the growth rate of earnings are both equal to zero.5
Retirement income includes only pension benefits. So while earned income is 15,000
euros until retirement, and zero afterwards, disposable income is 11,250 before retirement and
10,000 afterwards. Given the large contributions and corresponding benefits, discretionary
saving is a tiny component of income (350 euros before retirement and –900 afterwards).
5
Miles (1999) also works out an example in which discretionary saving is positive, while mandatory saving and
total saving are large and negative during retirement.
19
Mandatory saving before retirement is 3,750 euros, while mandatory dissaving during
retirement amounts to the pension received, –10,000. Even though life-cycle savings are
4,100 before retirement and –10,900 after, in this example the large swings in life cycle
saving are almost eliminated if one uses the conventional definition of income and saving. If
the contribution rate is raised to 27.3 percent, discretionary saving would be zero through life.
But it would be a mistake to conclude that the flat saving profile contrasts with the predictions
of the life-cycle model, while the consumer in fact follows exactly that model!
In more general settings, mandated savings affect the path of discretionary saving,
although the impact might not be one-for-one, as in the previous example. However, allowing
for different returns on discretionary and pension wealth, individual income growth,
imbalances in the social security system, and life uncertainty does not change the qualitative
insight of the example: where pension wealth is a major component of total wealth, the path
of discretionary saving is a very poor indicator of saving targeted for retirement.6
Some authors have applied this framework to microeconomic data and have adjusted
age-saving profiles including back pension contributions in income and subtracting benefits
(Bosworth, Burtless and Sabelhaus, 1991; Gokhale, Kotlikoff and Sabelhaus, 1996; Jappelli
and Modigliani, 1998). Since a large component of wealth is annuitized, these studies show
that the elderly decumulate substantial amounts of wealth during retirement.
The authors of the country-chapters recognize the importance of estimating pension
wealth for the dynamics of saving. In several countries (Germany, Italy, Japan, and the
Netherlands) they also include an estimate of the age-profile of pension wealth. However,
with the exception of the Italian study, when it comes to saving data no paper explicitly
recognizes that in order to estimate retirement saving one should subtract consumption from
earned income (not from disposable income). This is a serious limitation of the project.
Relying on the conventional definition of saving implicitly assumes that households are
myopic, and that they do not realize that contributions during the working-span entitle them to
receiving benefits when old.
2. Measurement and estimation problems
The papers are very careful in drawing attention to a variety of measurement problems
and definitions. Here I just want to point out two issues that should deserve some attention in
future research.
6
This remains true not only when mandated saving is fully funded, but also when the system operates as pay-asyou-go. It poses, however, additional measurement problems. When the system is funded, the rate of return on
contributions is primarily linked to market returns. In pay-as-you-go regimes, the return depends on income
growth, demographic variables and expectations of future reforms and is therefore more difficult to evaluate.
20
2.1 The return on wealth
In reality the returns on discretionary and pension wealth are not equal to zero, as in
the example above. Estimating these returns with microeconomic data poses difficult
problems. As pointed out by Borsch-Supan (2001), income from real and financial wealth
does not include capital gains (or losses), so saving figures computed as difference between
income and consumption can differ dramatically from those based on first difference of
wealth. A related problem that is often neglected is that measured disposable income includes
the nominal return from financial assets, not the real return.
In time series studies of saving it is quite common to subtract from income the
inflation-induced capital losses on financial assets and to adjust saving for inflation. To my
knowledge, there has been no attempt to adjust nominal returns for inflation in
microeconomic data. The adjustment is potentially important, because the average return on
the household portfolio depends on the returns on each of the assets in the portfolio. In turn,
there is ample evidence that participation in specific assets is strongly related to age (Guiso,
Haliassos and Jappelli, 2001). Thus, the inflation adjustment might impact the age-profile of
saving. Furthermore, the adjustment depends on the inflation rate itself. Since the latter differs
widely between countries, it is potentially important to explain patterns of saving across
countries.
The return on the social security component of annuitized wealth depends, to say the
least, on the growth rate of the economy, survival probabilities, legislation, and inflation. If
the system is balanced, the return on social security is equal to the real growth rate of the
economy, while survival probabilities increase the effective rate at which people discount
future wealth. If the system is not balanced, it involves intergenerational transfers. In the
terminology of Gokhale et al (1996), there is an "old-age tax" that is levied on current
generations that needs to be imputed. One of the major achievements of the International
Project is to report comparable figures for social security wealth over the life-cycle that take
into account some of the difficult problems in estimating internal rates of return for pension
wealth.
21
2.2 Saving rates or saving levels?
The life-cycle model makes clear predictions about saving levels, not about saving
rates as conventionally measured. In the stripped-down model with zero interest rates, income
earned during retirement is zero, and the saving rate is not even defined.7 With positive
interest rate, the consumer earns capital income during retirement, and the rate should be
negative, going to minus infinity at the end of life. In international comparisons there are
clearly some advantages in normalizing saving to obtain saving rates. Ideally the
normalization should use a variable that does not depend on age. Some authors (Attanasio,
1994) have proposed to normalize by consumption or permanent income. Even though also
these variables depend on age, this alternative is more appealing.
Most of the papers of the International Project, however, have chosen to focus on
saving rates as conventionally measured. Using repeated cross-sectional data, they decompose
these rates into cohort, age and time effects. The latter are restricted to be orthogonal to a time
trend and to sum to zero, and capture therefore only business cycle variations in saving rates.
While for variables such as income, consumption and wealth this is a reasonable identification
assumption (and very much in the spirit of the life-cycle model that assumes that productivity
growth is generation-specific), it is less appealing for saving rates. Paxson (1996) shows that
age effects in saving rates can differ if one uses alternative decompositions. For instance,
explaining saving rates by age and unrestricted time effects, assuming no cohort effects (i.e.,
that cohort effects in consumption and permanent income are the same) has a dramatic impact
on the age profile of US and UK saving rates. It would be very useful to complement the
evidence of the country papers with a check on the sensitivity of the age-saving profile with
respect to different identification assumptions. 8
3. The evidence
The International Project broadly confirms the set of findings in Poterba (1994): if
anything, there is little decumulation of private wealth, and discretionary saving is positive
even in old age. The first column of Table 1 summarizes some of the findings of the project. It
7
It is defined if one uses the conventional definition of saving.
8
It should also be pointed out that the identification assumption assumes that age, cohort and time
effects are additive. Several recent pension reforms, however, impact differently households of
specific cohorts, so that the presence of time-age-cohort interaction terms cannot be easily dismissed,
see Miniaci, Grant and Weber (2001) for an application to the Italian pension reform.
22
reports that in virtually all nations saving is positive well beyond retirement. In the United
States, Italy, the Netherlands, and Japan saving declines with age but, with the exception of
Japan, it is still positive at old ages. In other countries (France, Germany) saving actually
increases with age. In some countries the pattern contrasts with the evidence on saving as
measured by the first difference of wealth. For instance, there is evidence that discretionary
health declines in the Netherlands, Italy, and the United States. So the picture for changes in
wealth is rather different than that for saving, showing some inconsistencies between wealth
and saving measures. Such inconsistencies should be resolved before one can draw firm
conclusions about the shape of the age-saving profile.
The summary table also reports indication of the composition of wealth around
retirement. In all countries annuitized wealth (in the form of contributions to pension funds or
to social security) is quite substantial. For instance, in Japan and in Italy social security wealth
is about twice as high as discretionary wealth. In the UK, Germany and the Netherlands it is
about 50 percent higher; in the US, according to Gokhale et al (1996), annuitized wealth is
about 40 percent of total wealth at retirement. Unfortunately, the data do not allow a complete
picture, because we lack data for France and for private pension funds in several countries.
The counterpart of this is that the income of the elderly is mainly income from
annuities (see in particular the careful evidence about the source of retirement income in the
US, Dutch and UK studies). Thus, given the high contribution rates to pension funds and
social security, in all countries the difference between earned income and disposable income
is particularly large (positive during the working span and negative during retirement). And a
proper account of saving would show that the elderly do decumulate assets in all countries, in
that the consumption of the elderly is financed in large part by annuities. To consider just one
example, let's consider the Netherlands, where median wealth at retirement is only 35,000
euros. This tiny stock of wealth is slowly depleted during retirement, but not drained down to
zero, perhaps to face unanticipated health expenses, the risk of longevity and bequests. But
the bulk of retirement consumption is financed by the build-up of private pension funds
covering 90 percent of the workers and of a generous pension system.
We can speculate about the reasons why such a large part of wealth is annuitized, so
that in modern societies a great deal of saving occurs through mandatory retirement plans,
such as social security and private pension funds, while dissaving occurs mainly through
annuitized wealth. But even if people cannot choose the amount of mandatory saving, they
can change discretionary saving in response to changes in mandatory saving (Gale, 1998).
23
And after all, the existence of mandatory saving programs and the widespread implementation
of retirement plans should be interpreted as the social approval of schemes designed to ensure
people with adequate reserves to be spend during retirement.
The papers in this project have chosen instead to focus primarily on the age-profile of
discretionary saving. The shape of this profile does not tell us if households decumulate
wealth during retirement, but conveys information on a distinct issue, i.e. the importance of
intergenerational transfers. Even though wealth trajectories per se cannot be taken as evidence
in favor or against bequest motives, the age-saving profile represents therefore a useful
summary statistic on the role of intergenerational transfers.
4. Agenda for the economics of saving in Europe
The International Project reports very important insights on saving in individual
countries. It also defines an agenda for future research on the economics of saving in Europe.
Without trying to be exhaustive, I would like to draw attention on three issues: (1) estimating
the relation between mandatory and discretionary saving; (2) the need for panel data; (3) the
role of annuitized medical expenses.
The first important topic for research is that we need comparable estimates for
mandatory saving across countries. But most importantly, accounting definitions of the sort
that I discussed in Section 2 do not tell us anything about the relation between discretionary
and mandatory saving (or discretionary wealth and pension wealth). This impact is crucial to
understanding the effect of current pension reforms that reduce mandatory saving and
increase discretionary saving. In principle, an increase in contributions and benefits should
tend to be offset by a reduction in discretionary saving, leaving total saving unaffected (the
Feldstein replacement effect). In practice, social security systems encourage early retirement,
which would tend to increase discretionary saving (Feldstein’s and Munnel’s induced
retirement effect). Furthermore, pension wealth is illiquid and the social security component
of that wealth is subject to political risk. So even though by definition total saving is the sum
of discretionary and mandatory saving, it is not true that an increase in mandatory saving will
necessarily reduce discretionary saving in proportion. Having comparable European figures
for the relation between mandatory saving and discretionary saving would represent a major
improvement of what we know about intertemporal choices.
The second issue on the agenda is that we need to go beyond the use of repeated crosssectional data to analyze wealth and saving trajectories. If we want to estimate the relation
between discretionary and mandatory wealth, we need to impute pension wealth. But the latter
24
depends heavily on family composition and labor market histories. While discretionary wealth
is a well-defined concept (as are household consumption and household income) that can be
easily aggregated in cohort data with standard assumptions, pension wealth refers to
individuals, not households. Several country-studies show that pension wealth is affected by
the labor supply of the household (for instance, one vs. two-income earners), the age-gap
between spouses, their expected retirement age, their employment status (employed vs. selfemployed) and so on. Only panel data with information on retirement transitions and labor
market histories of individual household members are able to measure the substitution
between mandatory and discretionary saving. The need for panel data is particularly important
at a time when the distribution of retirement ages and pension benefits is changing due to
ongoing reforms.
Finally, annuitization of future medical expenses is not included in any measure of
saving of the sort that I have discussed. The presence of a national health system in each of
the countries analyzed poses important definition and economic issues. As with pension
contributions, payments to health insurance are also a form of mandatory saving and National
Health Systems, as any pay-as-you-go system, transfer resources from the young to the old
generations. As noted in the UK study, it is hard to impute these transfers to households (or
individuals), because contributions are often paid out the general income tax and benefits are
provided in-kind. Still, this is an exercise that is worth attempting, because differences in
health insurance arrangements could explain different patterns of consumption by the elderly
across Europe. As with the International Project, this is an area where collaboration between
experts of different countries will help us making progress in understanding saving behavior.
25
References
Attanasio, Orazio P. (1994) "Personal saving in the United States," in James Poterba (ed.),
International comparison of personal saving. Chicago: The University of Chicago
Press.
Borsch-Supan (2001), "International comparison of household saving behavior: a study of
life-cycle savings in seven countries," Research in Economics, this issue.
Feldstein, Martin (1976), "Social security and saving: the extended life-cycle theory,"
American Economic Review 66, 77-87.
Gale, William G. (1998), “The effects of pension wealth on household wealth: a reevaluation
of theory and evidence,” Journal of Political Economy 106, 706-23.
Gokhale, Jagadeesh, Laurence J. Kotlikoff and John Sabelhaus (1996), "Understanding the
postwar decline in US saving: a cohort analysis," Brookings Papers on Economic
Activity 1, 315-390.
Guiso, Luigi, Michalis Haliassos and Tullio Jappelli (2001), Household Portfolios.
Cambridge: MIT Press (forthcoming).
Jappelli, Tullio and Franco Modigliani (1998), "The age-saving profile and the life-cycle
hypothesis," University of Salerno, CSEF Working Paper n. 9,
http://www.dise.unisa.it/WP/wp9.pdf
Miles, David (1999), "Modeling the impact of demographic change upon the economy,"
Economic Journal 109, 1-36.
Miniaci, Raffaele and Guglielmo Weber (1999), "The Italian recession of 1993: aggregate
implications of microeconomic evidence", Review of Economics and Statistics 81, 23749.
Modigliani, Franco (1986), "Life-cycle, individual thrift, and the wealth of nations,"
American Economic Review 76, 297-313.
Paxson, Christina (1996), "Saving and growth: evidence from micro data," European
Economic Review 40, 255-88.
Poterba, James (1994), "Introduction," in James Poterba (ed.), International comparison of
personal saving. Chicago: The University of Chicago Press.
26
International Comparison of Household Savings Behavior:
A Study of Life-Cycle Savings in Seven Countries
Rejoinder to Tullio Jappelli’s Comments
Axel Börsch-Supan
Tullio Jappelli’s thoughtful comments very much enrich the set of the seven country studies
in the International Savings Comparison Project (ISCP). Let me focus this brief rejoinder on
two points: Our disagreement over the most appropriate saving definition, and our agreement
on the urgent need for better data to make progress.
Jappelli’s comments reveal a rather deep rift in what economists should define as “saving”:
Should it include mandatory contributions to unfunded pension schemes as Jappelli urges, or
should saving be defined narrower without these contributions as it is done conventionally
and by the ISCP. I personally think that this rift is more semantics and taste than substance,
but it sheds a very illuminating light on the center piece of this project.
There are many reasons to save. These days, it is fashionable to distinguish between low and
high frequency saving and their underlying long and short run motives. Very low frequency
saving is for retirement. In many countries, unfunded pension schemes (almost) completely
take care of retirement income. Hence, sticking with the conventional definition of saving,
there is no need for retirement saving and only higher frequency saving motives remain. The
data collected in the ISCP show that some of these motives persist after retirement since we
observe little or no dissaving in old age. More general, an important – maybe the most
important – insight from the papers in the ISCP is the significance of saving motives other
than retirement saving. Adding contributions to unfunded pension schemes to discretionary
saving dilutes this insight, and the only gain is a reconstruction of the textbook life-cycle asset
profile by a now tautological accumulation and decumulation of unfunded wealth.
Of course and as always, how one defines a measure depends on the purpose. If one wants to
show the validity of Modigliani’s honored life-cycle hypothesis of saving and consumption,
one needs to include contributions to unfunded pension schemes because doing otherwise
would leave out the main mechanism to smooth consumption between work and retirement.
This is Jappelli’s well-taken point. But we are beyond this point. Quibbling about the validity
27
of the life-cycle hypothesis using narrow saving definitions is a waste of time. We rather need
to understand (at least) two fundamental questions: Is the replacement between funded and
unfunded saving mechanisms one to one? And how do we explain the importance of high
frequency saving? Neither of these questions will be answered by including contributions to
unfunded pension schemes in saving. Hence, the narrow definition of saving employed by the
ISCP is not a “shortcoming” of the project, as Jappelli puts it, but a prerequisite to understand
these two important questions.
The importance of, and the interaction between, these two questions leads me to this
rejoinder’s second point in which I fully agree with Jappelli: we need more work to make
progress, and this progress requires more data. The ISCP has to do more work on the first
question by being more precise on the measures of pension generosity and taking care of all
the other determinants of saving – in order to establish a ceteris paribus comparison between
unfunded pensions and funded saving. This requires a large set of comparable covariates
which the current country data sets are short of. And while the second question has already
found many answers – precautionary saving, buffer-stocks, behavioral aspects –we step into
subtle terrain requiring better data in order to assign each motive its correct weight in total
saving.
From physics, we know that entering subtle terrain requires more precise observations.
Physicists need large machines to find small particles. Physicists are fairly successful in
finding funds to pay for these machines, and they have delivered impressive insights in the
make-up of matter. They were able to distinguish between ever more subtle alternative
explanations how matter is made up.
In economics, we are far behind. Economists, who want to gain insights in the make-up of
human behavior, at least an important aspect of it, need more (and more costly) data if they
want to get beyond answering simple questions. If we want to distinguish between saving
motives which are just a bit more subtle than the retirement saving motive (and even there if
we want to understand the substitution quantitatively), we need longitudinal data. If we want
to understand saving in old age, we need to have data on health along with more precise
measurements of asset decumulation. If we want to understand the precautionary savings
motive, we need to collect data on perceived risks. If we want to understand the transfer
motive of saving, we need data on the family network. If we want to understand the influence
of capital gains on saving, we need data on the individual portfolio, including housing
appreciation. And since not a single of these questions can be answered in isolation because
of potential substitution, we need data that shed light on all these aspects jointly.
28
Thus, an important lesson of both the Poterba exercise and the ISCP is that economists need
to join forces and put more effort into data collection. The American Health and Retirement
Survey is a great first step in this direction. Europe is much behind. I appreciate that Tullio
Jappelli’s comments have put a finger into this open sore of economic science.
29
Table 1
Findings of the International Project
Except where indicated, the data are taken from the papers of the International Project. All values are
expressed in euros.
Country
Discretionary
saving during
retirement
Evidence on
decumulation of
discretionary
wealth
Real plus
financial wealth
around
retirement
Pension wealth
around
retirement
Social security
wealth around
retirement
France
Increasing with age
200,000
n.a.
n.a.
(between 3,000 and
4,000)
No evidence of
decumulation
(age profile of
wealth is rather
flat after age 70)
Germany
Mean saving is
about 3,500, median
is 1,000 and both
are rather flat after
age 65
No evidence of
wealth
decumulation
(profile is flat
after age 60)
150,000
n.a. but
modest;
pension funds
finance only
6% of
retirement
income
225,000
Italy
Declining from
about 2,000 but still
positive in old age
Evidence of
wealth
decumulation
(Modigliani and
Jappelli, 1998)
90,000
n.a. but
modest given
limited role of
pension funds
180,000
Negative saving
rates after age 60 or
70, depending on
definitions
n.a.
128,000
Netherlands
Low and declining,
but still positive in
old age
Evidence of
wealth
decumulation
Average wealth
at age 60-64 is
83,000, median
is 35,000
n.a. but very
important
(90% of the
workers are
covered)
Between 100,000
and 177,000,
depending on
household
composition
United
Kingdom
Median saving after
age 65 is about
1,000
n.a.
n.a.
n.a. but very
important
(87% of the
workers are
covered)
109,200
United
States
Median saving
declines with age
but remains positive
Financial wealth
declines with
age
Japan
n.a.
(including the
severance pay
fund)
Between 215,000
to 320,000,
depending on
assumptions
about the internal
rate of return
(about as
important as
social security)
According to Gokhale at al (1996), for males in the 6069 age-group 41 percent of total wealth is annuitized;
for females in the same age-group 46 percent of wealth
is annuitized
30
The German Savings Puzzle
Axel Börsch-Supan, Anette Reil-Held,
Ralf Rodepeter, Reinhold Schnabel, Joachim Winter
MEA, Department of Economics and SFB 504, University of Mannheim, Germany
Abstract
Germany has one of the most generous public pension and health insurance systems of
the world, yet private savings are high until old age. Savings remain positive in old age,
even for most low income households. How can we explain what we might want to term
the “German savings puzzle”?
We provide a complicated answer that combines historical facts with capital market
imperfections, housing, tax and pension policies. The first part of the paper describes
how German households save, based on a synthetic panel of four cross sections of the
German Income and Expenditure Survey ("Einkommens- und Verbrauchsstichproben")
collected between 1978 and 1993. The second part links saving behavior with public
policy, notably tax and pension policy.
Address:
MEA, Department of Economics
University of Mannheim
D-68131 Mannheim, Germany
Email: [email protected]
This paper has profited greatly from discussions with Agar Brugiavini, Mike Hurd, Tullio Jappelli, Jim Smith
and Gert Wagner. We are grateful to Florian Heiss, Simone Kohnz, Melanie Lührmann, and Gerit MeyerHubbert for their able research assistance. We appreciate the financial support by the Deutsche
Forschungsgemeinschaft (Sonderforschungsbereich 504) and by the EU (TMR-Project „Savings, Pensions and
Portfolio Choice“).
31
The German Savings Puzzle
by Axel Börsch-Supan, Anette Reil-Held, Ralf Rodepeter, Reinhold Schnabel, and
Joachim Winter
• Introduction
This paper describes how German households save and why the observed savings patterns might
have emerged. In the descriptive part of the paper, we present cross sectional and longitudinal
patterns of household saving. We then explain why these saving patterns have likely been
strongly influenced by public policies. These policies include capital taxation and subsidies to
specific forms of saving, and, most notably, pension policies.
We face a “German savings puzzle”: Germany has one of the most generous public pension and
health insurance systems of the world, yet private savings are high until old age.9 We provide a
complicated answer to the questions raised by that puzzle, combining historical facts with capital
market imperfections, housing, tax and pension policies.
The paper is a brief version of the German country chapter in Börsch-Supan (2001). The reader is
referred to this volume for details on methodology and results. This summary paper is set up as
follows: Section 1 briefly describes our data sources. Section 2 presents cross-sectional and
longitudinal profiles of various saving measures by age and birth cohort. Section 3 looks at
financial, real and pension wealth. Section 4 links the observed saving and wealth patterns to
public policy, and Section 5 concludes.
• 1. Data
We base our description of savings behavior in Germany on four cross sections of the German
Income and Expenditure Survey ("Einkommens- und Verbrauchsstichproben," EVS). The EVS is
collected every five years by the German Bureau of the Census.10
9
The design roughly
A survey of the German pension system is provided by Börsch-Supan and Schnabel (1998).
Descriptive analyses of household wealth have been carried out by the German Bureau of the Census (Euler, 1985,
1990; Guttmann, 1995). The 1978-1988 surveys have been analyzed with respect to household savings by BörschSupan and Stahl (1991), Velling (1991), Lang (1998), and Börsch-Supan (1992, 1994).
10
corresponds to that of the U.S. Consumer Expenditure Survey. The surveys include a very
detailed account of income by source, consumption by type, saving flows, and asset stocks by
portfolio category. As opposed to earlier waves, the 1993 wave also includes households in East
Germany, and foreign residents in West Germany. For comparability reasons, we will restrict our
analysis to the subsample of West Germans.
The surveys are representative of all households with annual gross incomes below DM 300,000.11
They include about 45,000 households in each wave.
These large sample sizes provide
sufficiently large cell sizes in each age category, even for old ages. The data exclude the very
wealthy households and the institutionalized population. The former represent about two percent
of households. For this reason, the data cannot be expected to add up to national accounting
figures. For example, aggregating household savings in the EVS 1983 yields a net private saving
rate of 12.3 percent while the corresponding figure reported by the Deutsche Bundesbank is 13.6
percent.12 Omission of the institutionalized is serious only among the very old. Although less
than four percent of all persons aged 65 and more in Germany are institutionalized, this
percentage increases rapidly with age and is estimated to be about 9.3 percent of all persons aged
80 and more. Elderly in institutions are more likely to have few assets and no savings, hence, we
probably overestimate the assets of elderly persons.
Households in the quinquennial EVS cross sections are not necessarily the same and cannot be
matched. It is therefore impossible to construct a panel of individuals. This would be most
desirable for the identification of life-cycle saving behavior and the separation of age and cohort
effects. Lacking longitudinal data on savings behavior in Germany, we resort to the construction
of a synthetic panel. We aggregate the cross sectional data into age categories and identify
adjacent age groups across waves. The large sample sizes are of considerable help for the
synthetic cohort approach because aggregation units can be defined sufficiently narrow to assure
homogeneity without a major loss of statistical precision.
11
More details of the multi-stage quota sample design can be found in the full version of this paper (Börsch-Supan,
Reil-Held, Rodepeter, Schnabel and Winter, 2001). It also contains a more detailed comparison between aggregate
and survey based saving measures.
12
This divergence is due to two differences in the base: The EVS omit the upper 2 percent of the income distribution
while the Bundesbank also includes non-profit organizations.
33
• 2. Saving by age and birth cohorts
The data permit two measurements of savings.13 The first measure is computed as the sum of
purchases of assets minus sales of assets. Changes in financial assets reported in the EVS are
deposits to and withdrawals from the various kinds of savings accounts; purchases and sales of
stocks and bonds; deposits to and withdrawals from dedicated savings accounts at building
societies ("Bausparkassen") which are an important savings component in Germany; and
contributions to life insurances and private pension plans minus payments received. New loans
are subtracted and repayments are added to net savings. Not reported are changes in cash and
checking accounts. Changes in real assets reported in the EVS are purchases and sales of real
estate and business partnerships. Not reliably reported are changes in durables (other than real
estate). Unrealized capital gains are unreported. To arrive at saving rates, household saving is
divided by disposable household income, consisting of labor, asset, and transfer income minus
taxes and social security contributions.
The second definition of saving is the residual of income minus consumption. We will show that
both definitions are very close on average although there is substantial discrepancy for some
households. A third definition, the difference between initial and end of period stocks of wealth,
cannot be computed from the data since stocks are measured only once in each wave. Following
the definitions in the introductory paper of this journal issue (Börsch-Supan, 2001), we
distinguish among discretionary saving, composed of real and financial saving, mandatory saving
to funded pension plans, and “notional saving”, the mandatory contributions to unfunded social
security systems.
• Discretionary Saving
Figure 1 shows mean total discretionary saving by age in the four cross sections 1978-1993. On
the vertical axis, amounts are given are in real terms, converted to 1999 Euro.14 On the horizontal
axis, we have age, generally in five year intervals. Each age category also represents a cohort,
13
See the full version of this paper (Börsch-Supan, Reil-Held, Rodepeter, Schnabel and Winter, 2001) for details on
data sources, definitions for the variables used in this paper, and a discussion of measurement problems together with
our preferred solutions. All measures have been defined to be strictly comparable across waves. Börsch-Supan
(2001) also provides an electronic appendix with all data in spreadsheet form.
14
All amounts were deflated using the German consumer price index.
34
and following points on one of the cross sectional lines drawn in Figure 1 compares households
that are simultaneously in different age categories and cohorts.
The shapes are roughly similar. Changes across years are far from a simple shift of each profile:
for the younger age groups, 1978 and 1993 were the years with the highest saving, while there is
less of a clear picture for the older ones. There are two main features. First, saving exhibits a
hump shape, reaching a peak at the age/cohort group around age 45. Second, saving remains
positive, even in old age.
Figure 1: Mean Discretionary Saving in 1978-1993
These features are astoundingly similar for all income groups except the lower income quarter of
the German households, see Figure 2. Median and mean saving have the same hump shape as
Figure 1, and remain positive for all age groups, except for the lower quartile.
Figure 2: Mean and Median Discretionary Saving in 1993
While Figures 1 and 2 were calculated as purchases minus sales of assets during one calendar
year, the EVS also permits the computation of a second savings measure, namely the residual
from subtracting all consumption expenditures from disposable income.15 Figure 3 depicts the
comparison of both measures and shows that our saving measure is robust. The figure also gives
an impression of the sampling error of our saving measure which is relatively small due to the
large cell sizes.
Figure 3: Mean Discretionary Saving by Two Different Definitions, 1993
The first measure is almost always within the 2σ-confidence bands of the second measure.16
Using confidence bands for both measures, the difference is not significant. This is an important
result as it strengthens the belief in the internal consistency of the data, even though there are
some large deviations between the two measures for a few households which are masked by the
averages depicted in Figure 3.
15
Disposable income is gross income minus direct taxes and contributions to mandatory social security systems.
Consumption expenditures are reported very detailed in the EVS, based on weekly diaries. For precise definitions see
Börsch-Supan, Reil-Held, Rodepeter, Schnabel and Winter (2001).
35
Figures 1-3 display cross-sectional variation across age/cohort-groups and do not identify lifecycle changes. In order to do understand life-cycle behavior, we need to follow households over
time. As pointed out in Section 1, we lack longitudinal data on savings in Germany and therefore
combine the data of the available four EVS cross-sections from 1978 to 1993 to a synthetic panel
of household groups. Figure 4 displays cohort-specific age savings profiles from this synthetic
panel under the identifying assumption that time effects are zero, starting on the left with the
youngest cohort in our data, born between 1954 and 1958, and proceeding to the oldest cohort,
born between 1909 and 1913.17 Saving increases until it reaches a peak in the age range 45-49,
then declines until the age group of the 65-69 old. It then remains essentially flat. As pointed out
before, saving remains positive even in old age.
Figure 4: Mean Discretionary Saving by Cohort
The life-cycle pattern in saving visible in Figure 4 has two components: the hump-shaped pattern
of disposable household income,18 and the relatively flat pattern of saving rates to which we turn
now.
Saving Rates
Because mean saving rates are very sensitive to changes in nominator and denominator, we focus
on the median and quartile saving rates in each age category. We only show the 1993 cross
section since the others have a very similar shape. Figure 5 shows that the age/cohort pattern is
rather stable across income quartiles. The differences (pronounced hump shape for the richer,
fairly flat for the poorer households) are thus mainly due to differences in income profiles. The
increase in saving rates in very old age is interesting. Remember, however, that the data only
covers households, not elderly in institutions. Thus, the sample selects those who are less likely
to dissave. A back-on-the-envelope calculation (Börsch-Supan, 1992) shows that this selection
effect by itself is unlikely to explain the high saving rates in old age, although a precise analysis
cannot be done without genuine longitudinal data.
Figure 5: Median Saving Rates, 1993 Cross Section
16
The bands are computed under the assumption that the quota sample can be treated as a random sample.
Identifying assumptions in genuine and synthetic panels (Deaton, 1985) are discussed by Brugiavini and Weber
(2001).
18
Displayed in Börsch-Supan, Reil-Held, Rodepeter, Schnabel and Winter (2001).
17
36
If we combine the data visible in Figure 5 with the other waves and disentangle age and cohort
effects, we obtain the life-cycle profiles of Figure 6. Saving rates are fairly stable and around
12% for all young and middle-aged groups until around age 45-49. They then decline and
stabilize around age 65-69, when they remain at about 4%.19
Figure 6: Median Saving Rates by Cohort
Composition of Saving: Real and Financial Saving
Real estate saving, depicted in Figure 7, mainly consists of purchases minus sales of owneroccupied housing, including a correction for upkeep and depreciation, and subtracting applicable
mortgage payments.20 Figure 7 shows the four cross sections of real saving, 1978-1993. Because
homeownership in Germany is only about 40 percent, much lower than in most other countries,
the median is mostly zero and not shown. The means depicted in Figure 7 quickly reach a sizable
magnitude for the age/cohort groups around age 35 and then decline. Mean real estate saving for
the older age groups has a very large variance – it is mainly imputed depreciation and illmeasured upkeep – and is omitted from Figure 7.
Figure 7: Mean Real Saving, 1978-1993
Financial saving is relatively flat between age 30 and 40, then reaches a peak between age 40 and
45.21 Figure 8 shows the median for all four cross sections, Figure 9 mean and medium for 1993.
The flat part is most likely be due to the slow build-up or even withdrawal of financial assets
during the ages when many households purchase a house.
Figure 8: Median Financial Saving, 1978-1993
Mean and median financial saving are very close. This is visible in Figure 9 which shows details
of financial saving in the 1993 wave. As mentioned in Section 1, our data excludes the upper two
19
The data suggests an increase for the 1988 wave for all older cohorts. We have no satisfactory explanation for this
effect, particularly, because the pension level decreased between 1983 and 1988.
20
Other real wealth is not well measured. For example, the EVS data do not permit a sensible measurement of
changes in wealth that is invested in business partnerships. This does affect only a few households significantly but
not the average. See Börsch-Supan, Reil-Held, Rodepeter, Schnabel and Winter (2001). We also do not have the
regional information necessary to impute capital gains in housing which were large in some places such as Munich.
21
Our measure of financial saving includes the conventional financial saving categories, includes consumer loans,
but excludes mortgages as well as capital gains or losses. Capital gains to the consumer have been small in Germany
37
percent of the income distribution and thus misses households that deviate considerably from the
mean. It is noteworthy that financial saving remains positive even for those households that are
age 70 and older.
Figure 9: Mean and Median Financial Saving in 1993
Mandatory Saving
Mandatory contributions to public funded pension plans are negligible in Germany. Only a
minority of civil servants are required to contribute a small percentage of their salary increases to
funds that are effectively invested in government bonds. The contributions amount to roughly
0.5% of salary.
Contributions to private pension plans are not negligible in Germany, but they are much smaller
than, e.g., in the Netherlands or in the Anglo-Saxon countries. Slightly more than 50% of workers
are covered by a firm pension at least part of their career, but these pensions are small and
provide only about 6 percent of total average retirement income. In many cases, these pension
plans are mandatory in the sense that they come as a package deal with the employment contract
and offer no opting-out possibility.
Because mandatory occupational pensions play such a small role in Germany, the related saving
flows have been subsumed in the discretionary saving category discussed earlier.
“Notional saving:” Mandatory contributions to pay-as-you-go systems
Germany has very large pay-as-you-go systems that finance old age and health care. Almost all
dependent employees and their employers must contribute to the German public retirement
insurance. As pointed out in the introductory paper (Börsch-Supan, 2001), these contributions are
not saving in a narrow sense. However, they are a functional equivalent of saving and thus a
potentially important determinant for discretionary saving. We will discuss this extensively in
Section 4.
relative to the UK and the US, see Börsch-Supan and Eymann (2000).
38
The contribution rate to the public retirement insurance is 19.3% of gross earnings during the
year 2000.22 In addition, an estimated 8.5% of gross earnings is levied indirectly via other taxes,
mainly V.A.T. and the new ecology tax. The contribution base for public pension contributions is
capped at about 1.8 times the average earnings. Opting out is impossible. High wage earners
therefore pay a lower percentage of their income into the pay-as-you-go system and receive a
correspondingly lower replacement rate. The contributions add up to a claim on public pensions
that is substantial when compared to actual financial and real wealth. We turn to this point in the
following section on wealth.
Other branches of the German social insurance system include health, long-term care, and
unemployment insurance. For the average worker, the contributions to these branches add up to
another 21 percent of gross income.23 For the public health and long-term care insurance, the tax
base is capped at about 1.6 times the average earnings. Workers above this threshold can opt out.
The contribution base for the unemployment insurance is capped at about 1.8 times the average
earnings. Opting out is impossible.
In sum, these social insurance contributions by far exceed discretionary savings for all dependent
employees below the earnings cap – about 85 percent of all workers.
• 3. Wealth by age and birth cohort
The EVS also provide data on the stocks of financial, real and total discretionary wealth in a
separate interview at the end of each survey year. We use these data to cross-check our findings
on saving flows and to obtain a picture of total resources at the disposal of a household when the
household reaches retirement.
Discretionary real and financial wealth
Figure 10 depicts total discretionary wealth, defined in accordance to the flow measure of
discretionary saving in Section 2, and arranged by cohorts using the synthetic panel approach
described earlier. It consists of gross financial and real wealth, minus outstanding consumer loans
22
More precisely: Gross earnings include net earnings, income taxes and the employee’s share (one half) of social
security contributions. Total labor compensation includes gross earnings as defined plus the second half of social
security contributions, the so-called employer’s share.
39
and mortgages.
We see that total discretionary wealth increases until late in life, and there is only a brief (and
statistically insignificant) indication of a flat episode for the 1909 and 1914 cohorts, and even
there the change between the first and the last observation is positive.
Figure 10: Mean Total Discretionary Wealth by Cohort
West German private households possessed an average total wealth of DM 245,000 (€ 122,000).
At the time of the head’s retirement an average German household owned around DM 275,000 (€
138,000) of total wealth in 1993. This is 12.5 times the annual public pension of an average
employee with 45 years of service in 1993 (net DM 22,000, € 11,000). The median wealth at that
age is DM 200,000, (€ 100,000) which is lower than the mean but still relatively high. Thus,
drawing down wealth could quite substantially contribute to consumption (Schnabel, 1999).
Nevertheless, accumulation of even more wealth in the form of financial wealth takes place on
average in old age, as was illustrated in the savings profiles presented earlier. This is a surprising
departure from the life-cycle hypothesis.
The largest part of total discretionary wealth is real estate, in particular owner-occupied housing,
compare Figures 11 and 12. For the group aged 30 to 59, real wealth amounts to 80 to 90 percent
of total wealth. Mean gross real wealth increased substantially from 1978 to 1993. A more
detailed analysis shows that this is mainly caused by an increase in homeownership from cohort
to cohort, while ownership rates remained essentially constant with increasing age after age 60
for any given cohort (Schnabel, 1999).
Figure 11: Mean Gross Real Estate Wealth, 1978–1993
Financial wealth increased by 38 percent between 1978 and 1993. This increase was mainly
caused by a wealth expansion of middle age classes. The expansion of financial wealth is striking
between 1988 and 1993. The reason is a large increase in securities ownership for all age classes.
Figure 12: Mean Gross Financial Wealth, 1978–1993
23
See previous footnote.
40
Pension wealth
The life-cycle pattern of discretionary wealth in Germany – almost always increasing, at most flat
– is in contrast to the hump shaped pattern of unfunded (“notional”) pension wealth that trivially
emerges from the sequence of first paying pension contributions and then receiving pension
benefits. Figure 13 shows how notional pension wealth builds up and is drawn down in a
synthetic life cycle. The representative worker underlying this simulation has an earnings history
of the average age-specific wage between ages 20 and 60, then retires at the average retirement
age and draws the statutory pension benefits. Notional pension wealth SSW at time t is then
computed as24
SSW(t) = (1+rho)*SSW(t-1) + contributions(t) – benefits(t)
where rho is the internal rate of return that equalizes the present value of contributions and
benefits for the above 40-year contribution history and a duration of benefits corresponding to
average life expectancy. At retirement, notional pension wealth of the representative worker is
about DM 400,000, 30 percent more than the sum of average financial and real wealth shown in
Figure 11. By definition, notional pension wealth is drawn down after age 60 and becomes
negative after age 78, average life expectancy, see Figure 13. In contrast, financial and real
wealth increases until age 70 for the 1919 cohort (see Figure 10), and increases between age 60
(65) and age 75 (80) for the 1914 (1909) cohort. This contrast is not by chance. Rather, it reflects
the influence of pension policies on discretionary saving. This is the main argument of the
following section.
Figure 13: Life-Cycle Build-up of Notional Pension Wealth
4. Saving Patterns and Public Policy
We can summarize the observed saving patterns of German households in the following three
points:
♦ Saving rates are high and stable until around age 45-49.
24
See Brugiavini and Weber (2001) for a discussion of this measure.
41
♦ Saving is lower but still positive even in old age. There is depreciation drawing down real
wealth, but virtually no signs of drawing down financial wealth.
♦ Until age 35, saving is mainly invested in owner-occupied housing, while it is mainly
financial saving at older ages.
These observations pose a host of questions: How can we explain a life-cycle profile of
discretionary household saving in Germany which is much flatter than, e.g., in the US?
Specifically: Why does saving remain positive in old age, even for most low income households?
And what explains the “German savings puzzle”, the puzzling fact that pensions and health
insurance are generous and likely to have large crowding out effects, yet German households
accumulate so much real and financial wealth and do not appear to draw it down?
We need a complicated answer to resolve this puzzle. We obviously need to distinguish between
the older and the younger generation because they appear to save for different purposes.
Moreover, our data on the flat and positive savings in old age only pertain to the cohorts born
before the 1930s; we do not yet know whether that pattern will also hold for the younger
generation.25 We then distinguish among three effects of public policies: effects on the level of
savings, essentially by crowding-out mechanisms mainly through social insurance; effects on the
life-cycle pattern of savings, flattening the age-savings profile; and effects on the portfolio
composition of savings, mainly through differential taxation.
Crowding-out effects of public pensions
We start with an analysis of the older generation in our data. Their members were born between
1910 and 1930 and they retired until about 1995 – this is today’s generation of German retirees.
Their current income is dominated by public pension income, much more so than in many other
countries, see Table 1:
Table 1: Retirement Income by Pillar (Percentages)
About 85% of retirement income stems from the public mandatory retirement insurance, and only
15% come from private sources such as funded firm pensions, individual retirement accounts and
25
We refer to the generation now aged between about age 30 and 50. There is also a third generation, the “really
young”, but we have little data on their saving and consumption habits.
42
other asset income, only a little remaining labor income and family transfers.
The international comparison in Table 1 suggests a strong substitution between the provision of
pay-as-you-go pensions and other income sources in old age. This crowding-out result is in line
with a careful time-series analysis of Kim (1992). He links changes in the retirement system to
the savings rate and shows that the German pay-as-you-go system has crowded out saving to a
significant extent. Cigno and Rosati (1996) confirm these findings but explain the crowding-out
effect unconventionally by repercussions on fertility rather than through the familiar channels
stressed by Feldstein (1974).
43
The crowding-out result as it pertains to current retirement income is also at odds with the fact
that Germany has such a high saving rate, and in particular, that German elderly have on average
real and financial wealth levels that suffice for about 10 years of their retirement income (cf.
Figure 10). This is of course the core of the “German savings puzzle”. We need three elements to
explain it.
First, a part of the apparent contradiction between stocks of wealth (almost equally divided
between notional pension wealth and tangible real and financial wealth) on the one hand and
current income (85% pensions, 15% other income) on the other hand is resolved by realizing that
Table 1 only reports current money income, not the imputed rent from homeownership, and that
most wealth held by the elderly is owner-occupied housing (cf. Figures 11 and 12). Hence, Table
1 exaggerates potential crowding-out effects. However, the omission of imputed rent cannot fully
explain the puzzle. The German homeownership rate is much lower than in the Netherlands, the
UK, and the US. For the generation born between 1910 and 1930, it is just above 50 percent. 26
Moreover, flat and positive saving rates in old age are also prevalent among elderly German
renters (cf. Figure 2).
Schnabel (1999) provides the second element of our explanation. It is a story of ex ante versus ex
post savings plans. He shows that the growth of income during the German economic miracle
years and up to the seventies was so large and unprecedented that the elderly could just not have
anticipated it. Hence, they saved more than if they had known how miraculous a growth rate they
would experience.
Figure 14 displays the growth of earnings during the work history of a typical worker who retired
in 1970, the drop due to the 70% replacement rate after retirement, and then the subsequent
increase in pension income due to gross indication. All numbers are in real terms. After less than
10 years into retirement, the average worker had essentially recouped the former income level.
The process was only stopped in the early eighties, when economic growth slowed down to
normal also in Germany. Since such an income path could hardly be anticipated, workers
consumed too little and ended up with too large a stock of wealth around retirement.
26
This lower homeownership rate is only partially offset by the fact that the average home in Germany is more
expensive than in the Netherlands, the UK, and the US, see below.
44
Figure 14: Life-Cycle Income Path of the 1910 Cohort
While Schnabel’s (1999) story is plausible, it does not explain why this wealth has not been spent
at higher rates in old age. This is the third element of our explanation of the “German savings
puzzle.” First, habit formation may play a role. The elderly do not want to change the accustomed
level of consumption which they have learned some 50 years ago, not even increase it in the face
of accumulated financial wealth. There is some new evidence on the importance of habit
formation (Dynan, 2000). Second, Börsch-Supan and Stahl (1991) provide a complementary
explanation. They argue that due to deteriorating health conditions, the elderly are less able to
spend as much as they would need to make saving negative. Both lines of argument are
strengthened by capital market imperfections since annuitized pension income cannot be
borrowed against. Hence, even if the current generation of elderly had anticipated their
unwillingness or inability to draw down wealth at later ages, they could not have responded by
dissaving faster as long as their annuity income exceeds the planned consumption level. Evidence
for this effect is provided by Börsch-Supan (1992).27
Life-cycle saving patterns
While the older generation may have had a retirement savings motive, but was surprised by the
high retirement income and could not draw the accumulated wealth down, the younger generation
– now aged between about 30 and 50 years – has learned that retirement will not be a time of
scarce resources. For them, the high replacement rates of the German public pension system have
made additional private retirement provision largely unnecessary. Saving for retirement, the only
motive under the pure life-cycle hypothesis, is of secondary importance. Other saving motives
dominate, most importantly saving for homeownership, as Figures 7 and 8 have shown. In
addition, there are motives such as high frequency precautionary saving, high frequency saving
for durables such as cars, and saving for intergenerational transfers. In fact, inter vivos transfers
are high in Germany and survey questions on savings motives show an almost equal spread
between the aforementioned saving motives (DIA, 1999).
27
We know very little about bequests which may, in theory, contribute to the observed flat age-saving profiles in old
age. Cross-section regressions of wealth levels on number of children do not produce significant results. This
finding, however, does not necessarily rule out an operative bequest motive. Only longitudinal data will clarify this
matter.
45
The mechanisms pertaining to both generations generate much flatter age-saving profiles than
under the retirement-saving oriented life-cycle hypothesis. The older generation still has positive
saving rates because of the unwillingness or inability to draw down wealth at later ages which
they have accumulated in lack of anticipation of the spectacular economic growth. The young
generation has a flat saving profile because the slow process of owning a home and shortfrequency saving motives generate a flat saving rate over a long period.28
Hence, the generous public pension system in Germany appears to be the main cause for a
relatively flat age-saving profile. It has made the retirement savings motive relatively irrelevant
for the younger generation, and it has led to overannuitization among the elderly. We are aware
that this line of argument is vulnerable because it lacks a counterfactual. The international
comparisons in this journal issue do help in this respect. For instance, among the countries
represented, the hump-shaped life-cycle savings pattern is most pronounced in the U.S. where the
replacement rate of the public pension systems is lower – and thus the retirement savings motive
is more important – than in continental Europe.
If a substantial portion of the saving patterns currently observed in Germany is caused by the
public pension system, we should expect substantial changes in saving patterns in the future.
Growth rates have declined and the dependency ratio is deteriorating rapidly. The current
generosity of the social insurance system is unlikely to prevail. A major pension reform is under
way which will cut benefits substantially and, in effect, introduces more prefunding. This will
revive the retirement motive for saving. Hence, saving rates among the young are likely to
increase, and saving rates among the elderly are likely to decline sharply because the have to rely
more on their retirement savings to fiance consumption. We will have to wait for this
counterfactual to obtain a clearer explanation of what caused the puzzling German savings
behavior.
Portfolio composition
Public policies appear also to have shaped the composition of tangible household wealth.29 As
pointed out in Section 3, the largest part is real estate, mainly owner-occupied housing. For the
group aged 30 to 59 this makes 80 to 90 percent of total wealth. While ownership rates are lower
28
Conventional mortgages in Germany have a term of 30 years.
46
than in most other European countries, the US and Japan, both land and housing construction is
relatively expensive in Germany. This paper is not the place to analyze why this is the case, but
there is some evidence pointing towards restrictive land regulation.30 In addition, saving for down
payment in building societies (“Bausparkassen”) is tax privileged.
Tax policy appears to have shaped the composition of financial wealth, displayed in Table 2.31
Table 2: Composition of Financial Household Wealth, 1978–1993
The most important component is whole life insurance, about a third of gross financial wealth.
The central reason for the important role of whole life insurance in German households life-cycle
savings decisions is its favorable tax treatment, as shown by Brunsbach and Lang (1998) and
Walliser and Winter (1999). Stocks and bonds are the second most important category. Bonds
make up the lions’ share in this category, while stocks are less than 10 percent of the average
household portfolio. This fact is also significant for financial markets, as life-insurance
companies have not been allowed to invest significantly in stocks in the past, which in turn is one
of the main reasons for thin capital markets in Germany. Stocks and bonds are tax privileged in
so far as capital gains are tax exempt if the underlying asset has been held for longer than one
year.32 The lenient taxation of capital income may be another explanation for the high saving rate
in Germany, but we are not aware of a reliable time series analysis that links the level of tax relief
to the aggregate household saving rate.
It is highly speculative how the portfolio composition in Table 2 would change in the wake of a
major change of the German social insurance system, notably a partial transition to prefunding
pensions. If there were no substitution between new retirement saving and current saving, the
household saving rate would increase by between 2 and 4 percent, see Birg and Börsch-Supan
(1999). If these new savings were channeled into pension funds, which only recently have been
introduced in Germany and still do not receive preferential tax treatment similar to whole life
29
For a detailed study of German household portfolio choice, see Börsch-Supan and Eymann (2000).
Börsch-Supan, Kanemoto and Stahl (2001) claim that housing policies explain a significant share of the price
differences among Germany, Japan and the US, such as restrictive land development by local governments,
excessive building codes and insufficient legislation to avoid monopolization of the construction industry.
31
A survey of tax policy in Germany is provided in the companion paper Börsch-Supan, Reil-Held, Rodepeter,
Schnabel and Winter (2001).
30
47
insurance, pension funds would amount to between 15 and 18 percent of households’ portfolios,
comparable to the United Kingdom, the U.S., the Netherlands and Switzerland. Substitution
between new retirement saving and current saving would increase this share, but part of new
retirement saving may also be done as whole life insurance. Households’ direct and indirect
exposure to stock markets then depends on future investment decisions of life insurance
companies who only recently began to increase their portfolio share of stocks. Judging from the
international experience in countries as diverse as the United Kingdom, the U.S., the Netherlands
and Switzerland, a more prominent role of equities seems very likely when more of the German
retirement income is prefunded.
5. Conclusions
The case of Germany presents an interesting “savings puzzle.” One the one hand, saving rates are
high and stable until around age 45-49, and remain positive even in old age. While depreciation
draws down real wealth among elderly homeowners, we find virtually no signs of drawing down
financial wealth. One the other hand, Germany has a very generous public pension system.
“Notional pension wealth” provided by the pay-as-you-go social insurance system is larger than
real wealth and much larger than financial wealth.
Our explanation is cohort-specific. Our data on the flat and positive savings in old age only
pertain to the cohorts born before the 1930s; we do not yet know whether that pattern will also
hold for the younger generation. The older generation was surprised by an unprecedented income
growth in the 1960s and 70s. Households born between 1910 and 1930 were saving for
retirement but ended up being over-annuitized. Habit formation and ill health then prevented the
older generation from spending their unexpected wealth down.
What will happen, when younger cohorts reach retirement, is likely to depend on future pension
policy. Pension reform is under way in Germany. It will shift a significant share – between a
quarter and a third – of retirement income from the pay-as-you-go pillar to a funded pillar. Most
likely, this will increase saving in younger ages, and induce dissaving among the elderly. We will
have to wait for this “experiment” to obtain a clearer explanation of what had caused the puzzling
German savings behavior.
32
This has recently been changed to two years.
48
References
Birg, H., and A. Börsch-Supan (1999): Für eine neue Aufgabenteilung zwischen gesetzlicher
und privater Altersversorgung, Berlin: GDV.
Börsch-Supan, A. (1992): Saving and consumption patterns of the elderly: the German case.
Journal of Population Economics, 5, 289-303.
Börsch-Supan, A. (1994): Savings in Germany - Part II: Behavior. In: Poterba, J. (ed.),
International Comparisons of Household Savings. Chicago, London: University of
Chicago Press, 207-236.
Börsch-Supan, A. (ed.) (2001): International Comparisons of Household Saving, New York:
Academic Press.
Börsch-Supan, A., and A. Eymann (2000): Household Portfolios in Germany, mimeo,
University of Mannheim. Forthcoming in: L. Guiso, M. Haliassos and T. Jappelli (eds.),
Household Portfolios, Cambridge, Mass.: MIT-Press.
Börsch-Supan, A., and A. Reil-Held (1998): Retirement income: level, risk, and substitution
among income components. OECD Ageing Working Paper AWP 3.7. Paris.
Börsch-Supan, A., and R. Schnabel (1998): Social Security and Declining Labor Force
Participation in Germany.’ American Economic Review 88.2, 173-178.
Börsch-Supan, A., and K. Stahl (1991): Life-cycle savings and consumption constraints.
Journal of Population Economics, 4, 233-255.
Börsch-Supan, A., Y. Kanemoto and K. Stahl (2001): Housing Markets in Germany, Japan
and the United States, unpublished manuscript, University of Mannheim.
Börsch-Supan, A., A. Reil-Held, R. Rodepeter, J. Winter and R. Schnabel (2001): Household
Saving in Germany. In: Börsch-Supan, A. (ed.), International Comparisons of Household
Saving, New York: Academic Press.
Brugiavini, A., and G. Weber (2001): Household Savings: Concepts and Measurement. In:
Börsch-Supan, A. (ed.), International Comparisons of Household Saving, New York:
Academic Press.
Brunsbach, S., and O. Lang (1998): Steuervorteile und die Rendite des
Lebensversicherungssparens, Jahrbücher für Nationalökonomie und Statistik 217, 185213.
Cigno, A., and F.C. Rosati (1996): Jointly Determined Saving and Fertility Behaviour: Theory,
and Estimates for Germany, Italy, UK, and USA, European Economic Review 40, 156189.
Deaton, A. (1985): Panel data from time-series of cross-sections. Journal of Econometrics, 30,
109-124.
Deutsche Bundesbank (1998): 50 Jahre Deutsche Mark: Monetäre Statistiken 1948-1997.
München: Verlage C.H. Beck, Vahlen.
Deutsches Institut für Altersvorsorge (DIA, 1999): Die Deutschen und Ihr Geld. Köln:
Deutsches Institut für Altersvorsorge.
49
Disney, R., M. Mira d’Ercole and P. Scherer (1998): Resources during retirement. OECD
Ageing Working Paper AWP 4.3. Paris.
Dynan, K. (2000): Habit Formation in Consumer Preferences: Evidence from Panel Data. The
American Economic Review 90, 391-406.
Euler, M. (1985): Geldvermögen privater Haushalte Ende 1983. Wirtschaft und Statistik, Heft 5,
408-418.
Euler, M. (1990): Geldvermögen und Schulden privater Haushalte Ende 1988, Wirtschaft und
Statistik, Heft 11, 798-808.
Guttmann, E. (1995): Geldvermögen und Schulden privater Haushalte Ende 1993. Wirtschaft
und Statistik, Heft 5, 391-399.
Kim, S., (1992): Gesetzliche Rentenversicherung und Ersparnisbildung der privaten Haushalte in
der Bundesrepublik Deutschland von 1962 bis 1988, Zeitschrift für die gesamte
Versicherungswirtschaft 81, 555-.
Lang, O. (1998): Steueranreize und Geldanlagen im Lebenszyklus: empirische Analysen zu Sparund Portfolioentscheidungen deutscher Privathaushalte. Baden-Baden: Nomos Verlag.
Reil-Held, A. (1999): Bequests and aggregate wealth accumulation in Germany. The Geneva
Papers on Risk and Insurance, 24, 50-63.
Schnabel, R. (1999): Ersparnis und Vermögen im Lebenszyklus in Westdeutschland.
Habilitation, University of Mannheim.
Velling, M. (1991): Ersparnisbildung in Westdeutschland. Diplomarbeit, University of
Mannheim.
Walliser, J. and J. Winter (1998): Tax incentives, bequest motives and the demand for life
insurance:
Evidence
from
Germany.
Discussion
Paper
No.
99-28,
Sonderforschungsbereich 504, University of Mannheim.
50
Figure 1: Mean Discretionary Saving in 1978-1993
7 00 0
6 00 0
Savings in Euro
5 00 0
4 00 0
3 00 0
19 93
19 88
19 83
19 78
2 00 0
1 00 0
0
20
30
40
50
60
70
80
A g e -g ro up
Note: All data in prices of 1993 and weighted. Age/Cohort-groups denoted by begin of 5-year interval.
Source: Own calculations on the basis of the EVS 1978–1993.
Figure 2: Mean and Median Discretionary Saving in 1993
12000
10000
Savings in Euro
8000
6000
75%
4000
Mean
2000
Median
25%
0
20
25
30
35
40
45
50
55
60
65
70
75
80
85
-2000
Age-group
Note: All data in prices of 1993 and weighted. Age/Cohort-groups denoted by begin of 5-year interval.
Source: Own calculations on the basis of the EVS 1978–1993.
51
Figure 3: Mean Discretionary Saving by Two Different Definitions, 1993
10 00 0
9 00 0
8 00 0
Savings in Euro
7 00 0
6 00 0
5 00 0
4 00 0
S 2-2s ig m
3 00 0
S2
2 00 0
S 2+ 2 sigm
S1
1 00 0
0
20
25
30
35
40
45
50
55
60
65
70
75
80
85
A g e-g ro u p
Note: S1 is the first measure (purchases minus sales of assets), shown with the 95%-confidence bands (S2±2σ). S2 is
the second measure (residual of income minus consumption). All data in prices of 1993. Age/Cohort-groups denoted
by begin of 5-year interval. Source: Own calculations on the basis of the EVS 1978–1993.
Figure 4: Mean Discretionary Saving by Cohort
8000
6000
Savings in Euro
1909
1914
1919
1924
1929
4000
1934
1939
1944
1949
1954
2000
0
20
25
30
35
40
45
50
55
60
65
70
75
80
85
A g e G ro up
Note: All data in prices of 1993. Age-groups denoted by begin of 5-year interval. Source: Schnabel (1999)
52
Figure 5: Median Saving Rates, 1993 Cross Section
30%
25%
20%
Saving Rate
15%
10%
75%
Mean
5%
Median
0%
20
25
30
35
40
45
50
55
60
65
70
75
80
85
25%
-5%
-10%
Age-group
Note: All data in prices of 1993. Saving is defined as purchases minus sales of assets. Age/Cohort-groups denoted by
begin of 5-year interval. Source: Own calculations on the basis of the EVS 1978–1993.
53
Figure 6: Median Saving Rates by Cohort
16%
14%
12%
1914
1919
1924
1929
1934
1939
1944
1949
Saving Rate
10%
8%
6%
4%
2%
0%
25
30
35
40
45
50
55
60
65
70
75
80
85
Age-group
Note: All data in prices of 1993. Age-groups denoted by begin of 5-year interval. Source: Schnabel (1999)
54
Figure 7: Mean Real Saving, 1978-1993
3500
3000
Real Savings in Euro
2500
1993
1988
1983
1978
2000
1500
1000
500
0
25
30
35
40
45
50
55
60
65
Age-group
Note: All data in prices of 1993 and weighted. Source: Own calculations on the basis of the EVS 1978–1993.
55
Figure 8: Median Financial Saving, 1978-1993
3500
3000
Financial Saving in Euro
2500
2000
1993
1988
1983
1978
1500
1000
500
0
20
25
30
35
40
45
50
55
60
65
70
75
80
Age-group
Note: All data in prices of 1993 and weighted. Age/Cohort-groups denoted by begin of 5-year interval.
Source: Own calculations on the basis of the EVS 1978–1993.
56
Figure 9: Mean and Median Financial Saving in 1993
8000
7000
6000
5000
Savings in Euro
4000
MEAN
MEDIAN
P25
3000
2000
P75
1000
0
18
21
25
30
35
40
45
50
55
60
63
66
70
75
80
85
-1000
-2000
-3000
Age-group
Note: All data in prices of 1993 and weighted. Age/Cohort-groups denoted by begin of 5-year interval.
Source: Own calculations on the basis of the EVS 1978–1993.
57
Figure 10: Mean Total Discretionary Wealth by Cohort
200000,00
Total Wealth in Euro
150000,00
1909
1914
1919
1924
1929
1934
1939
1944
1949
1954
100000,00
50000,00
0,00
20
25
30
35
40
45
50
55
60
65
70
75
80
85
Ag e-g rou p
Note: All data in prices of 1993. Age-groups denoted by begin of 5-year interval. Source: Schnabel (1999)
58
Figure 11: Mean Gross Real Estate Wealth, 1978–1993
160.000
140.000
Real Estate Wealth in Euro
120.000
100.000
80.000
1.993
60.000
1.988
40.000
1.983
1.978
20.000
0
25
30
35
40
45
50
55
60
65
70
75
80
85
Age-Group
Note: All data in prices of 1993 and weighted. Age/Cohort-groups denoted by begin of 5-year interval.
Source: Own calculations on the basis of the EVS 1978–1993.
59
Figure 12: Mean Gross Financial Wealth, 1978–1993
60.000
Gross Financial Wealth in Euro
50.000
40.000
30.000
1993
1988
1983
1978
20.000
10.000
0
25
30
35
40
45
50
55
60
65
70
75
80
85
Age-Group
Note: All data in prices of 1993. Age/Cohort-groups denoted by begin of 5-year interval. Source: Own calculations
on the basis of the EVS 1978–1993.
60
Figure 13: Life-Cycle Build-up of Notional Pension Wealth
250
Notional Pension Wealth (1000 Euro)
200
150
100
50
0
20 22 24 26 28 30 32 34 36 38 40 42 44 46 48 50 52 54 56 58 60 62 64 66 68 70 72 74 76 78 80 82
-50
-100
Age-Group
Note: All data in prices of 1993. Source: Own calculations, based on the average earner in the EVS 1993.
61
Figure 14: Life-Cycle Income Path of the 1910 Cohort
13000
12000
Earnings and Pensions in Euro
11000
10000
9000
8000
7000
6000
5000
4000
3000
2000
1920
1930
1940
1950
1960
1970
1980
1990
Source: Schnabel (1999)
62
Figure 15: Household Saving Rates in Germany
18 %
16 %
Household Saving Rate
14 %
12 %
10 %
G e rm a n y (W e st O n ly )
G e rm a n y (A ll)
8%
6%
4%
2%
0%
19 50
19 60
1 970
1 980
199 0
200 0
Y ear
Source: Deutsche Bundesbank (1998).
63
Table 1: Retirement Income by Pillar (Percentages)
Germany
The Netherlands
Switzerland
UK
US
State
85%
50%
42%
65%
45%
Employer
5%
40%
32%
25%
13%
Individual
10%
10%
26%
10%
42%
Notes: Income composition of two-person households with at least one retired person. UK: „State“ includes SERPS.
US: „Individual“ includes 25% earnings, much less in the other countries.
Source: Börsch-Supan and Reil-Held (1998) and Disney, d’Ercole and Scherer (1998).
Table 2: Composition of Financial Household Wealth, 1978–1993
1978
1983
1988
1993
Share in
1993
Savings accounts
8.721
6.863
7.459
6.243
17.5%
Building societies
3.495
3.344
2.806
2.663
7.5%
Bonds and stocksa
4.171
5.028
5.828
11.199
31.4%
Life insurance (cash value)
9.386
9.443
12.564
11.869
33.3%
Other financial wealth
-
1.017
1.002
3.713
10.4%
Gross financial wealth
25.773
25.695
29.659
35.687
100.0%
./. Loans
12.936
16
16.991
19.680
Net financial wealth
12.837
9.494
12.667
16.007
Note: Household data from the Einkommens- and Verbrauchsstichprobe (EVS). All figures in DM and in 1993
prices. a) About 70% bonds and 30% stocks. For details see Börsch-Supan, Reil-Held, Rodepeter, Schnabel and
Winter (2001).
64